First author

Humans and monkeys can distinguish between faces in a fraction of a second, seemingly without conscious effort. The neuronal mechanisms of recognition are unclear, so, while working at the Max Planck Institute for Biological Cybernetics in Tübingen, Germany, David Leopold investigated them (see page 572). Using trained monkeys, he and his colleagues monitored individual neuronal responses in a part of the brain involved in visual recognition while the monkeys viewed and responded to composite images. Leopold, who is now at the US National Institutes of Health in Bethesda, Maryland, tells Nature about this work.

Why take on this topic?

Because facial recognition is a remarkable, swift and effortless process. It allows us to rapidly navigate a wide variety of social situations, and is important for both survival and reproduction.

What was it like watching the monkeys responding to the images?

The role of the neurons was not clear during the experiments; the monkeys were simply viewing a series of randomly ordered faces on a monitor, then moving levers to select which faces most resembled each other. Afterwards, we put the faces in order and compared the variations in facial identity with the neuronal signals. This was exciting, because we realized that single neurons in this part of the brain respond in a highly systematic way to very small variations in facial characteristics.

What do the monkeys' response speeds say about the nature of recognition?

Their responses were almost mechanistic, and so fast that the monkeys wouldn't have had much time to think about which face was which.

What is the importance of the 'average' face in this paper?

It seems that single neurons in our brains compare new faces with an internally stored 'prototype' face, which is the average of faces we have seen before. This is important, because for us to understand face recognition, we have to consider that the neurons aren't encoding all the details of a face; they are encoding the deviations from a stored reference. This average face is something you or the monkey has accrued with experience.

What's next?

We'd like to look at the neuronal basis of learning in primates, and how the average neuronal representation of faces and other complex objects comes about.